Vestibular function after cochlear implantation is difficult to understand, as subjective vestibular symptoms seem uncorrelated with the results of objective tests. Consequently, clinicians may struggle to decide what assessments to perform for a symptomatic patient. We used a systematic review and meta-analysis approach to enlighten this point. After a study inclusion process, results were classified into four different groups for each test in each study: (1) 'true positive' if the test showed impairment from pre-operative to post-operative in symptomatic patients; (2) 'false positive' if the test showed impairment from pre-operative to post-operative in asymptomatic patients; (3) 'true negative' if the test showed no impairment in asymptomatic patients; and (4) 'false negative' if the test showed no impairment in symptomatic patients. From these groups, sensitivities and specificities of each test were calculated in a meta-analysis. After reviewing more than 3000 references, 16 studies were included, representing 957 patients. The meta-analysis revealed a sensitivity of 0.21 (CI 95 % 0.08-0.40) for the caloric tests, of 0.32 (CI 95 % 0.15-0.54) for the cervical vestibular evoked myogenic potentials (c-VEMP), and of 0.5 (CI 95 % 0.07-0.93) for the head impulse tests. The analysis of prevalence revealed that c-VEMPs were the most often impaired, and the HIT the most often conserved. Our review and meta-analysis revealed that no vestibular test is sensitive enough to be recommended as a single test. Ideally, all the five vestibular sensors should be tested. In clinical practice, we suggest a case-to-case strategy according to patient's symptoms and their suspected origin.
Keywords: Cochlear implantation; Dizziness; Meta-analysis; Systematic review; Vertigo; Vestibular function test.